基于模型结构和事件日志的流程相似度计算

Calculating Similarity Between Business Process Based on Model Structure and Event Log

摘要:

流程相似度的计算在企业业务流程管理中具有重要作用。目前相似度的计算主要存在两个问题：一是大多数相似度计算方法只考虑模型结构或事件日志,导致算法不够精确；二是综合考虑了模型结构和事件日志的算法复杂度高且效率低。因此,提出了一种改进的流程模型结构和事件日志相结合的方法。首先将流程模型结构中的紧邻活动转化为邻接矩阵,然后根据事件日志中的行为信息对邻接矩阵进行加权得到加权邻接矩阵,最后采用符合距离度量特性的矩阵间距离的算法来度量流程间相似度。通过实验与MDS、GED以及WBPG等算法进行对比,所提方法的准确率更高,为99.51%,计算效率也更高。

Abstract:

The calculating of business process similarity plays an important role in enterprise business process management. At present, there are two main problems in the calculation of similarity: one is that most methods only consider the model structure or event log, which results in inaccurate algorithm, the other is that the algorithm considering the model structure and event log has high complexity and low efficiency. An improved approach is proposed to calculate similarity by combining model structure and event log. Firstly, constructs the adjacency matrix based on the adjacent activities of process model structure. Then the weight adjacent matrix is obtained by weighting the adjacent activities according to the behavior information in the event log. Finally, the inter-matrix distance algorithm conforming to the distance metric is used to measure the business process similarity. By comparing with the algorithms such as matrix distance similarity (MDS), graph edit distance (GED) and weight business process graph (WBPG), the accuracy of the proposed approach is 99.51%, and the calculation efficiency is higher.